Shared publicly  - 
 
Another year of AOSP

Well, the work year is over for me as today was my last day in the office before the new year.

This closes my 3rd whole year working on the Android Open-Source Project (my involvement in 2008 had been quite minor). It's been an interesting year for AOSP, full of contrasts.

On the positive side:

-I finally managed to distribute some proprietary hardware-related binaries for the very first time. This is an issue that I had been working on since before the G1 even shipped, and it finally became reality. For me personally, that's really a huge deal, because it really makes it practical to take the AOSP code, modify it, install it on a phone in a way that actually works, and to distribute the resulting image. This enables community builds of AOSP in ways that were not possible before, thanks to a license that allows redistribution.

-I also finally managed to distribute some factory images for the very first time. Just like for the proprietary binaries, this is the result of 3 years of work. The main benefit is obvious: play with an AOSP build for while, and then get your device back to its factory state, over and over and over. It also has some other nice advantages, e.g. it allows AOSP users to stay up-to-date with the latest bootloader and baseband firmware.

-For the first time, AOSP supports a non-Nexus non-ADP device, which also happens to be a tablet (Xoom), and supports a development board (PandaBoard). We also added support for Intel CPUs, including emulator support. Those show well that AOSP is important for companies other than Google.

-We also got some proprietary binaries submitted directly in AOSP instead of being separate downloads, e.g. the touchscreen firmware for Galaxy Nexus and the Wifi/Bluetooth firmware for PandaBoard.

-We reached the point where AOSP is running on 11 different targets (8 different devices from 4 different families, plus 3 virtualized or emulated systems).

-IceCreamSandwich was a massive release, which added over 220,000 changes over Gingerbread. As usual since Froyo, the full source history is available, so that people can easily study how the code evolved, or research when, why and by whom a given line of code was written.

-With IceCreamSandwich, we now have a fully Open-Source NFC stack and a fully Open-Source sensor processing chain. In Galaxy Nexus, we also have a wifi/bluetooth chip that can work without requiring any proprietary firmware.

-I tagged 26 device releases in 2011, compared to 12 releases in 2010.

-I fine-tuned the process that we use when releasing GPL source files, so that it's now much easier to build the GPL source code independently from full platform releases.

-We Open-Sourced the source.android.com site itself, thanks to the work on the intern who took care of that in the beginning of the year.

-Last but not least, the AOSP team grew, and that allows us to split the workload: +Conley Owens takes care of the long-term engineering while I manage the day-to-day execution. Conley gets credit for implementing the auto-verifier that we had running on Gerrit for a while, for dealing with many build issues that appear on various environments, and also for a lot of other work that's not visible yet but that'll make things even better in the future. All that happens under the watchful eye of +Dan Morrill who occasionally stops by to say hi, and with a lot of help from Google's main Open-Source team.

On the negative side:

-Not releasing the Honeycomb source code was catastrophic for the AOSP community. I had never before received so many angry emails, so many threats, to the point where I had to take several weeks off at some point to get away from it. Even today, there's a lot of bitterness left on all sides. From start to finish, Honeycomb probably cost AOSP anywhere from 6 to 12 months.

-We had 7 weeks of downtime on our Git hosting, and 4 months (and counting) on Gerrit. The Git outage was manageable as it happened at a time of low activity, but it still consumed a lot of engineering effort. However, the Gerrit outage couldn't have come at a worse time: not having Gerrit right as IceCreamSandwich was getting released means that we lost a precious opportunity to merge in many external contributions at a point where AOSP and Google's internal master branch were very close, the closest in about 2 years. No conspiracy here, just some really bad luck.

-In spite of a lot of progress, there are still far too many proprietary binaries and factory images that I'm not allowed to distribute, and I'm still wasting far too much time trying to convince those many companies that they should be helping AOSP run on their hardware (well, I've given up on HTC, Motorola and Qualcomm, as it's clear at this point that they're not going to help in that area).

Overall, in spite of a few serious difficulties, it's been another good year for AOSP.
350
134
JF Dionne's profile photoJoshua Talley's profile photoGreg Clarke's profile photoKeyan Mobli's profile photo
101 comments
 
Sounds like you had a good year! Nice work! (Thanks from the PandaBoard team!)
 
Good deal... I think 2012 will be a banner year!
 
Hopefully Motorola will be more helpful when they're fully inside the tent pissing out than they are now, outside it pissing in.
 
I think one of the things I love most about Android, is being able to see the REAL people putting REAL passion into their work for the benefit of us all. I for one, am glad to be aboard. Keep up the great work!
 
Great work as always, now give us 4.0.3 OTA as a New Years present :)
 
+Scott McLaughlin - I'm not the one making decisions on OTAs, that's in the domain of retail products, not Open-Source, so it's all managed under totally different constraints (including and especially operator approvals).
 
I've said it before, and I guess I should say it again... is there any way we can show direct appreciation for the AOSP work? Like, a beer fund or something? :)
 
One would expect that, with Motorola being acquired by Google, things in that regard will/would get better...One can only hope.
 
Ahhh, thanks for taking the time to answer my silly question. Would love to see the official 4.0.3 image hit the site sooner rather than later :)
 
+Richard Cohen - nah, I'm all set with alcohol (Germain Robin XO at the moment). Occasionally, I'd appreciate if people could show a tiny bit more patience, but then I also try to remember that all the pressure happens because people really care, i.e. because AOSP matters.
 
Thank you a lot Jean-Baptiste. Some great progress in 2011 and I hope 2012 will be greater! AOSP rocks!!
 
Closed drivers craze me. The secrets worth keeping should be (and usually already are) in the silicon. How an OS talks to your silicon should be easily discussed / improved in the open. Manufacturers have nothing to lose and everything to gain with open-source drivers.

My deepest thanks for the AOSP and all who make it possible. Exciting times.
 
+Scott McLaughlin - Yup. The two are tied. It doesn't take much work to put the image online, and there's no need for complex approvals, so unless I'm on vacation there's no reason why there'd be a huge delay after the OTA. Plus, as long as there's an older image, you can flash that and get the OTA (I know it's not very elegant, but it gets the job done).
 
One meta-thing to mention... there are a lot of companies where a post like this from an engineer would be severely frowned upon - particular the stuff about the HC screw-up!
 
+Ron Waldon - We're getting there for just about everything except OpenGL, which is the area where I don't expect anything to move for quite a while. No situation is ever as black-and-white as you suggest, though. We've seen an Open-Source GPS implementation already, and I've got my fingers crossed that we might see an Open-Source GSM RIL eventually (no hope for CDMA, though, that's far too proprietary).
 
I know it's not much, but all your work and stresses and patience is greatly appreciated. Thanks for helping to make Android something worth supporting!
 
+Richard Cohen - It wouldn't do anyone any good to pretend that Honeycomb didn't happen and that Android jumped straight from 2.3.2 to 4.0.1.
 
It always baffles me that in this day and age hardware companies continue to not recognize the usefulness of Open Source. It means getting software improvements and support for no additional cost. If the manufacturers would release their code then there are plenty of people who would be willing to clean out the bugs and give tech support when they no longer want to. +1 to +Jean-Baptiste Queru and the rest of the guys at AOSP for supporting the cause.
 
+Joshua Hull - I think that most companies recognize the value, but also find out that there are some significant costs associated with Open-Sourcing code. There's the direct cost of scrubbing the code clean. There's the direct cost of confirming the origin of the code. There's thedirect cost of managing regulatory compliance. There's the indirect cost of letting your competitors create new products while you spend time managing all those other costs. Those costs are especially hard to manage when it's so hard to find qualified engineers. There are also plenty of risks related to patents that I'm not even going to mention in detail here, but the situation is not pretty at all. There's also the risk that your system's security relies on secrecy that can be maintained in Open-Source code.

As part of Google's work on Android, we're finding that it's generally easier to re-write hardware-related code from scratch in a way that can be Open-Sourced rather than to try to get existing proprietary code Open-Sourced.
 
Not expecting Motorola to play ball in the new year? Isn't there somebody that can give them a call once things on the merger front are official?
 
+Jean-Baptiste Queru - You raise an excellent point about why if you are going to open source code it needs to be a decision that is made at the beginning of a project. Otherwise the cost of open sourcing can out weight the benifits. Unfortuantly I think we can all see how patents can harm open source but that's best left for the lawyers to fight over.
 
+Aaron Berlin - I'd rather be pleasantly surprised with no effort from my side than sink a lot more effort into that direction for nothing. There are plenty of other companies that are more willing to work with AOSP and I'd rather give those companies my attention.
 
Thanks for all your hard work +Jean-Baptiste Queru ! My next phone is going to be non HTC specifically because they insist on using sense. probably stock with the nexus line.
 
+Joshua Hull - Yes, it's very much easier to write code in order to Open-Source it than to Open-Source code that wasn't written that way.

As for patents, I think it's a bit naive to think that those are lawyer-only issues. Patents involve engineers at just about every level, and pretending that the patent system doesn't exist won't make it go away. It's especially important in the Open-Source world.
 
+Jean-Baptiste Queru Thanks for all your hard work. I have learned lots from following your work on AOSP and your posts. I am beginning to get into building for devices from source. I have successfully built ICS for the emulator in Ubuntu 11.10. Lots to learn :-)
 
+Jean-Baptiste Queru - My apologies, what I meant was that I didn't want to draw others into a fight over the issue. This thread is about celebrating the work of the Open Source community and you guys at AOSP, not bumming people out. But I agree, the patent system is something that needs to be looked at and not ignored.
 
Thanks for your work.
I wish the gerrit to come back soon.
 
+Joshua Hull - Thanks :) AOSP is pulling into uncharted territories (there are no other similarly successful Open-Source phone platforms), and that makes companies discover new challenges and new possibilities. Things have been generally positive, 2011 opened many doors, and I feel that many more doors will open if I knock a little bit more.
 
+SeongJae Park - Indeed, we all want Gerrit to be back as soon as possible, and the team at Google that manages our Git solutions has been putting all their efforts behind Gerrit so that we can use it for AOSP again as quickly as possible.
 
+Jean-Baptiste Queru Thanks for everything! Android has been a huge part of my life over the past couple years, so I want to thank you for your dedication. I've begun to start working with the CyanogenMod source code recently, and it's become...perhaps...an addiction? And, without you (and AOSP), the CyanogenMod project would not exist.

The power of open source is simply amazing, just take a look at the CM source tree. CyanogenMod supports (on the 'gingerbread' branch, at least) over 50 targets (devices) officially, with many more unofficially supported by individuals, built upon the base CyanogenMod source. All of these targets are built from the same source tree, which is downright amazing.

Yeah, perhaps HC was a snafu, but we all make mistakes. And it's not a bad thing; we live and learn. I understand why it was done -- when it comes down to business, you've got to do what you've got to do.

I love watching AOSP evolve, and I look forward to seeing the direction of it throughout 2012 (and beyond!). I've been a long-time fan of Nexus devices, owning two Nexus Ones and a Nexus S -- my fiancee had to comprimise on a LG/T-Mobile G2x this time around, though, due to an issue with T-Mobile's backend systems, and she's not too happy about it.

Our next devices will be a shiny pair of Galaxy Nexii (Nexuses?). I'm curious to see if the GSM version will land (semi-)officially in the USA, one way or another. If not, I'll be importing them. Yeah, I'm that dedicated. (And I hate CDMA. And not to mention the fact that VZW's LTE implementation is...buggy...at best.)

But, all in all, thanks! Keep up the good work!
 
Am I correct or just insanely naive to think that the problem with Honeycomb was that it was just a hack job developed under the pressure to market a tablet-friendly OS? If true, I wouldn't have released the code either because of how difficult it would be for developers to manage it, not to mention the embarrassment in releasing sub par work to the public.
 
I can't wait for your "another year of answering everyone questions" post. You are a man amongst boys here! Thanks JQB
 
HTC, Motorola and Qualcomm. Got it. Although I will probably only buy nexus devices from here on out (be it phone or tablet), I now know to make an effort to avoid whatever these companies sell until they change their tune with regard to AOSP. Thank you JBQ
 
Looking after the Android framework for one of the top 5 smartphone chip suppliers, there is nothing more than I and my colleagues would love to see than to have our HAL and BSP packages fully open sourced as part of the AOSP.

Where this falls apart is that at the business end, we lack a carrot with which to really implement this goal - such an agenda has to be supported and signed off by us, our customers / suppliers and partners as a unit and this is no mean feat ;0(

The group of companies that are involved in a single smartphone platform remain bonded together through a series of "non-disclosure agreements" and "software development agreements" - dragging everyone along onto the AOSP party bus would require more lawyers going out to lunch than we have seats in the best bistros... Even if we split the check, the bean counters will ask why and as always, its tough to sell good will.

What I believe we are missing is a stick from which the partners in a smartphone platform can be beaten with. If Google were propose that companies that behave as 'first class AOSP citizens' will get their just rewards, then I can see the business development guys from the respective companies getting excited and forcing the issue through the pipeline. For example, actively supporting a device in AOSP over multiple Android revisions gives you in exchange, early access to future release development branches prior to main open sourcing milestone (under the old NDA of course!) - this would be pretty juicy and hard to ignore I feel...

FYI, I have also been working on the AOSP for 3 years - Dec 2008 was the first baseband port we did for CES and time has flown by. Its been really fun on the other side of the fence - ICS is probably the best release to date - good job all involved on it!
 
Thanks for your work. Can we have some details on companies supporting AOSP?
 
So if google gets the go ahead on the Motorola deal, do you think then we will see the Motorola asop thing get better?
 
Just adding a huge thanks to the Android and AOSP teams, pushing OSS software like this is just amazing :) Also thanks to you +Jean-Baptiste Queru for speaking out on AOSP releases, updates and thoughts in general :)
 
+Jean-Baptiste Queru You mentioned that Motorola, Qualcomm and HTC won't let you distribute binaries. Are there any companies that seem likely to do that?
 
You are doing a great job. I especially like to thank you for releasing the GPL kernel sources in a timely fashion.
 
+Nick Lambourne Unfortunately, in my experience, your company is not all too FOSS friendly. I will say, I'd love to see that change. Glad to hear someone on the inside shares a similar view. :)
 
Thanks for the awesome efforts!

Will gitweb also be coming back? I have references to android.git.korg URLs throughout my source code, bug trackers, code reviews etc. It would be nice to have gitweb back, and some mod_rewrite magic from korg to googlesource.com (or wherever).
 
So Who's inbox should we spam to get them to allow distribution of proprietary binaries?
 
+Ankit Aggarwal - The Honeycomb code was as good as any other release, if there'd been something to hide we wouldn't have included it in the history of ICS, but since it's all out there people can go have a look. The problem was that the UI on phones ended up being a mix-and-match of Honeycomb-style UIs squeezed into a phone screen, Gingerbread-style UIs that didn't quite work on top of code that had been modified for Honeycomb, all around a phone system UI that didn't implement all the Honeycomb APIs. That was simply for lack of time: we developed Gingerbread and Honeycomb on the same schedule, and even though all the Gingerbread changes were all automatically merged into Honeycomb in real-time until surprisingly late (January 29), there simply was no time left after Gingerbread to create a phone UI in the Honeycomb timeframe, and that's where ICS came into play. The build date codes tell the whole story: 2.3.3 was the "true" Gingerbread, with the full Gingerbread API (level 10), and it was built on February 9 (build GRI40), while the original factory image for the Xoom was built on February 8 (build HRI39) with API level 11.

So, solid code in Honeycomb, with a UI that hadn't been tuned for phones. In a world where most people don't see deeper than the UI, there'd have been too big a cognitive dissonance between the real quality of Honeycomb (good) and the perceived quality (horrible). At the scheduling level, if we had instead not made any UI changes in Gingerbread, those same people who don't see deeper than the UI wouldn't have understood that Gingerbread contained a lot of important improvements in depth, especially for games, and if we had simply not done Gingerbread at all the devices without the power to run Honeycomb (or now ICS) would have been stuck on Froyo and would never have seen those same improvements.
 
+Nick Lambourne - I can't comment on other aspects of what's going on in your company, but from where I'm sitting there'd been no difficulty getting a redistribution license from your side for the WiFi/Bluetooth chip in Nexus S and in Xoom (in fact that was the earliest such license that we got for Xoom), and I very much appreciated to discover that the WiFi/Bluetooth chip from your company that is included in Galaxy Nexus works without any proprietary firmware. So, everything isn't sharply black-and-white, and there are some nuances in the way each company supports Open-Source.
 
+Adam Sobotka - There are a fair number of companies beyond Google that support AOSP, and I'm probably going to forget some. Without a doubt, Intel, Sony-Ericsson and TI should be mentioned in that list (that's alphabetical order). Imagination Technologies and NVIDIA were very friendly in letting me distribute their proprietary graphics libraries, each time getting all the necessary contracts ready on time. Special mentions go to Invensense (for the sensor code), Melfas (for the Galaxy Nexus touchscreen firmware), NXP (for the NFC stack). Samsung has been mostly helpful overall, in spite of not really understanding AOSP. Paradoxically, Qualcomm has also been helping in specific areas, especially around Gerrit and the build system.
 
+Jean-Baptiste Queru , +Nick Lambourne As far as wifi (firmware and drivers) is concerned, and even contributions to bluez, patchram, and some other BT friends, I don't think anybody can really complain.
I'd love to see BTL-A opensourced someday, though :)
 
+Mahfooz Hasan - Yes, there are definitely quite some companies that let me distribute proprietary binaries: Imagination Technologies, TI, Melfas, NVIDIA, Broadcom, Samsung, NXP, AKM.
 
+Steve Donaghy - No need to spam anyone. Doing so tends to alienate the best friends you have in such companies. The only real result in doing that is getting people angry and unmotivated. I continue to be in touch with those companies, I'm not letting them off the hook yet.
 
+Christopher Orr - We're looking into it, but right now the top priority in that domain is to get Gerrit running, so we've put gitweb on the back burner for now.
 
+Jean-Baptiste Queru , +Steve Donaghy Not just that, it also introduces a lot of noise into the discussions. People mean well, but they frequently ask for the wrong things; Talking to vendors, I've probably spent more time clarifying or even denying requests than actually making them :-(
 
Absolutely, I should have perhaps made it more obvious but I was speaking figuratively. What I really mean is, what's the best way to try and convince companies to be a bit more open? I mean, would a petition or something similar help? A greater show of support for those that have been open? Ultimately we are a community and so if there's anything the community can do together, then it should be worth considering.
 
+Steve Donaghy - surprisingly, one of the most important things is to be reasonable and patient. If the community sends an image of being demanding, picky, noisy, companies will simply not want to deal with anyone at all. The worst you can do is to appear uncompromising, to show that you'll only ever be satisfied with the purest of the pure Open-Source Free Software Copyleft: if you send a message that you won't see any value in any of the intermediate steps that a company might make, they won't even bother making the first step. Also, using those companies' property without having a license or working around the restrictions set by those companies isn't seen with a good eye: if you show that you're not willing to play by the rules, they won't want to play at all.

Every time a company hears demands that all their code should be released under the GPL, or reads bad press about alleged GPL violations, or sees people bypassing locked bootloaders, distributing unlicensed proprietary files, RMA-ing overclocked phones that overheated, it gets that much harder to convince that company to do the right thing, as it already takes a lot of effort just to get back to a point where such a company doesn't see Open-Source as the enemy. In any such discussion, both sides need to understand one another. The companies need to understand what's important for the Open-Source community, but the Open-Source community also needs to understand what's important for companies. No company wants to be associated with people who create bad press, and no company wants to be associated with people who don't respect IP rights.
 
+Jean-Baptiste Queru Happy Holidays and Happy New Year from Jamaica! Respect for doing what do you man!
 
Couldnt agree more with your com/HTC comments, hence my switch to the galaxy nexus. I still have my g1, and plan on installing RC28 to it to see just how much things have changed. Thanks to all you and everyone else do
 
+Keyan Mobli - Running 1.0? Now that's brave. Looking all the way back in a single step and realizing how much progress there's been in only 3 years causes a big shock. Be prepared!

I'd tell you to download the 1.0 source code and try to build it, but I haven't actually restored it on the new server. Amusing factoid: 1.0 had fewer than 100 git projects.
 
Just started playing with the android source and be involved with the "building" android community a couple of months ago, but still, I can see how passionate you are about this project. Thanks a lot for taking the time to answer all questions asked by the community.

Je te souhaite une belle année 2011 et longue vie au projet android !
 
JBQ, thank you for all you do!!! It's sad that they shot the messenger for Honeycomb but CM would be nowhere without you and my Nexus S would be nowhere without both you and CM. I hope you have a wonderful New Year!!!
 
Funny, that you've mengioned Motorola. It's a Google's property, isn't it? :)
 
+Oleg Gryb - Motorola Mobility is an independent publicly traded company (MMI), and a partner company with which Google developed some flagship devices: Droid and Xoom. My relationship with them is on that basis only, and if they continue to not be interested in helping AOSP there's nothing more I can do.
 
JBQ - first and foremost - thank you for all of your hard work. Android is a big win and AOSP and folks like you are why.
 
I know this is about 2011 in review, but what do you look most forward to in 2012 for AOSP? :)
 
+Al McElmon - Oh, interesting question. Here are some off-hand:

#1 - Gerrit coming back, so that we can get external contributions flowing in again.

#2 - A full set of distributable proprietary binaries for Galaxy Nexus (I've got my fingers crossed).

#3 - Using PandaBoard as a starting point for people to help make Android work better on TVs and computers.

Bonus points:

#4 - Improving the handling of proprietary binaries.

#5 - Running out of the box on actual Intel-based hardware (and not just emulated or virtualized).
 
I'm going to be buying three phones in 2012. And probably recommending phones to a few more people. If +HTC's ICS phones are supported by AOSP, that's who i'll be buying from, since i love how well the N1 has held up for two years. And thank you +Jean-Baptiste Queru for not just your work on AOSP, but also for keeping us well informed via the google group and g+.
 
What are ZTE, HTC, Motorola, Qualcomm etc; hiding in their driver source - are they so poorly written that they're not up to peer review, or do they hide bugs in the hardware that are patched with software? Or is it all eavesdropping/backdoors?

AOSP on non-Galaxy devices isn't going to go much further when we're trying to fudge proprietary blobs from Eclair onto ICS, even companies like Archos, HTC or even Raspberry who open things up a bit, are still crippled with closed blobs from Qualcomm.

Even getting Android vendors to comply with the GPL and release their modifications to the Linux kernel is hard enough, getting proprietary drivers out of them must be a nightmare.

What we need is a truly open Android phone/tablet/htpc, using no Qualcomm graphics/wifi drivers, opensource RIL, sensors, touchscreen etc.
 
To begin with I want to thank +Jean-Baptiste Queru and his ASOP colleagues for all the wonderful work. If it wasn't for ASOP then I wouldn't have the privilege of being able to follow projects such as CM, Replicant (http://www.replicant.us) and The Guardian Project (https://guardianproject.info).

Also, I will always buy and recommend hardware that follows the FOSS spirit the closest. Currently my options seem to be limited to the Nexus-line; I hope the handset/silicon manufactures can take a hint. Other than that, the previous commentators have said it all.
 
OK, with that we /know/ the bad carriers and whom to avoid; which companies are shining examples for AOSP?
 
+Avuton Olrich - I think you'll have to scroll back a little bit to see my earlier comments where I give quite a few examples. For me, the "shining examples" would be Intel, Sony-Ericsson, TI, though there are so many companies involved that I must be forgetting some.
 
+Simon John - I don't think that they fundamentally have anything they need to hide. The practical reality is that this is code that has been written organically over many year, since those companies have been in the cell phone business for extremely long. It's really hard to Open-Source such code after the fact. One common issue is that such code might implement protocols that aren't publicly documented anywhere else, that might have been developed by or in collaboration with another company, and figuring out whether it's possible to Open-Source an implementation of such a protocol is really hard. It's much easier to Open-Source code that was explicitly written with Open-Sourcing in mind, because those issues can get resolve as people go (and in fact before the code even gets written), but doing it after the fact is much harder.

Personally, I'm not hugely enthusiastic about the idea of putting so much effort behind re-writing hardware-related bits simply because there currently exists no Open-Source version. I guess it makes me feel that people don't care about the work that I do to Open-Source the Android platform itself. I also think that it's a waste of effort, since the result is by definition hardware-specific, so it won't help AOSP run on new hardware, and in the unlikely event that the end result does work better than the existing redistributable binaries this is still probably not the best way to improve the end-user experience with that amount of effort. Oh well, that's an opinion, but I still think that the Open-Source portable parts of Android are far enough from perfect that improving those would provide the best bang for the buck.
 
It's indeed like a roller coaster for Android@2011, I just hope it won't recur in the future.
 
+Jean-Baptiste Queru Thank you for believing in yourself and the open source communities, using all your energy,time and efforts to struggle to make this come true. I among others look up to you for leadership, patience and wisdom. Hopefully this becomes a reality one day and we can celebrate this achievement as a united community... Thank you for being you!
 
+Jean-Baptiste Queru I agree that I don't see the point in rewriting the closed hardware drivers in an opensource manner, but that does leave us to the mercy of vendors when trying to port to a device.

For instance how do we get libEGL.so which is ripped from GB with a quasi-2.6.32/35 kernel to work nicely with ICS without the help of the vendor? Without that working ICS is not a nice experience.

I think its going to have to boil down to a list of chipset vendors who are open like Intel/TI and closed like Qualcomm, but its going to be fun picking a phone based on every single chip in it, as well as bootloader shenanigans. Maybe we need an "AOSP approved" list which hopefully won't only consist of Galaxy phones that you've fought to get opened.
 
+Simon John - When it comes to graphics, we have to assume that the official libraries will remain proprietary for a while, that the effort to re-write such libraries independently is so huge that it is impractical, and that doing so without bumping into any patents is probably impossible. Also, even if you had an Open-Source Gingerbread graphics library, the effort to make it work with ICS would also be major.

In the end, when it comes to consumer devices, "Nexus" is where it's at. Google's influence on such device includes making them suitable for AOSP (including the way we pick specific chips and specific silicon vendors). It's not perfect, but there's progress each time, and I'm already thinking ahead to make things even better, both for the existing devices and for future ones.
 
as previously mention, thank you ! A huge thanks to you and your team for the great effort this year ! Looking forward to what you guys pull off in 2012

And thanks for sharing this cap of 2011 :) Happy New Year
 
+Jean-Baptiste Queru, I'm sorry to hear that that is how you feel. I sure that none of us thought that the lack of fully open hardware-specific software in any way diminished the effort put in by the ASOP team.

You raise some very valid points and unfortunately the reality of things do not always line up with ones hopes. Yet, the way I see it the Android ecosystem is a mix of software and hardware. There are heaps of reasons[1] for wanting open hardware-specific software and if the manufacturers cannot see the benefits or rationalize the (in)direct costs of writing such software (at the given moment) then they could at least I) stop trying to hinder people experimenting with hardware they legally purchased, or even II) provide data sheets for the silicon[2], and III) give FOSS some thought at the next iteration of R&D.

Anyhow, happy new year to you all! :-)

1: Guaranteed forward and backwards compatibility, educational purposes or slight paranoia.
2: For some a waste of effort, for others an educational journey in hardware/software interaction.
 
+Fernando Miguel, I must admit I have very limited insight in the history of Openmoko but I think it's fair so say that times have changed. As you've noted yourself, Linux 3.3 should be a great step in the world of Android.
 
+Tanel Rebane - In my experience working with hardware manufacturers and silicon vendors, they very much understand the benefits of Open-Source, but they also see the costs, and in the highly-competitive resource-constrained environments they operate in the costs include the opportunity costs of diverting engineers from creating new products.

It's important to keep in mind that at the silicon level the product cycles are very long, and 5 years is a good rule of thumb. If they made a decision a year or two ago to make Open-Source a high priority, it'll still take 3 or 4 years before that has an impact on a shipping product. Also, a big part of the difficulty is that development is iterative: when building on something that's not originally Open-Sourceable, the result ends up not being Open-Sourceable either, and the only way to break the loop is to re-start from scratch, which is essentially never an option given the fast pace of the industry.
 
+Jean-Baptiste Queru, this could be avoided if they help with documentation to some OpenSource developers or the community, but the hw industry moves really fast...
BTW Happy New Year
 
+Rafael Campos - Indeed, but writing documentation is more expensive than writing code (it takes more effort to test documentation than to test code). Also, if there's some magic sauce, it can possibly be implemented in code in a way that doesn't tend to stick out, but documentation would have to explicitly mention it.
 
+Jean-Baptiste Queru Thanks. I was asking because Princeton is still blocking Android devices from its network en masse, and the issue is still marked as "New" on the Google Code page I linked to.
 
+Chetan Reddy - I didn't learn anything new. While at a consumer level all variants of Galaxy Nexus are similar, in the case of AOSP I have confirmed that only the ones that were originally sold with a yakju build are expected to work.
 
I know this post is now quite old, but anyway, about this:
"-In spite of a lot of progress, there are still far too many proprietary binaries and factory images that I'm not allowed to distribute, and I'm still wasting far too much time trying to convince those many companies that they should be helping AOSP run on their hardware (well, I've given up on HTC, Motorola and Qualcomm, as it's clear at this point that they're not going to help in that area)."
will you try to negotiate with Motorola again, it being "under new management" ?
Add a comment...